-
Notifications
You must be signed in to change notification settings - Fork 26
Build Hangs at "Creating an optimized production build" on Multi-Node Cluster #80
Description
Description
I am experiencing an issue where a Next.js (Turbopack) deployment consistently hangs/fails at the final build stage: Creating an optimized production build .... My setup consists of a multi-node cluster managed via QuickStack/Longhorn.
Current Environment
Node A (Master/Small): ~48GB Disk, 4 ram. Currently showing high resource usage.
Node B (Worker/Large): 352GB Disk, 6 ram. Mostly empty.
Platform: QuickStack
Storage: Longhorn v1.7.2
The Issue
Build Failure: The deployment reaches the npm run build stage but terminates or hangs indefinitely during the optimization phase. No specific error code is given, suggesting an OOM (Out of Memory) kill by the OS.
Scheduling: Despite having a second, much larger node available, the build job appears to be pinned to the smaller, resource-constrained node.
Storage Conflict: Initially, the small node was marked "Unschedulable" due to high replica counts (defaulting to 4). Even after reducing replicas to 2 and freeing disk space, the build still fails.
Steps to Reproduce
Connect two VPS nodes of unequal size.
Attempt to deploy a standard Next.js project using next build --turbopack.
Observe the build log output until it reaches the optimization step.
Expected Behavior
The build should either:
Successfully complete by utilizing available RAM.
Automatically schedule the Build Job onto the node with the most available resources (the larger node).
Logs
Plaintext
[2026-04-02T22:11:30.999Z] #8 1.908 Creating an optimized production build ...
[2026-04-02T22:14:53.248Z]: [END] Log stream ended for build process: build-app-frontend-...